AbstractEquations were derived for predicting starch and lipid content of pea varieties from near infrared data measured on a filter instrument (Technicon InfraAlyzer 400). The accuracy of the predictions were tested against separate data sets and were found to be better than ±2.6% for starch (range 20–45%) and ±0.3% for lipid (range 1.4–3.3%). Together with the previous work on the near infrared determination of protein it is possible to analyse rapidly for three components important to a plant breeding programme. The study has also demonstrated a clear‐cut delineation between round and wrinkled pea ge